Objective
The project gathers an international network of seven institutes to investigate the possibility of electrochemical control of silicon surfaces for electronic and solar applications. The main goal is the elucidation of the mechanisms governing the anodic oxidation and dissolution of silicon, and the influence of these processes on the surface characteristics. The work is organised around the three following objectives :
- Characterisation of the oxide films present at the silicon surface during electropolishing or after anodic oxidation;
- Identification of intermediate chemical products and determination of reaction pathways in the anodic dissolution of silicon;
- Assessment of the role of crystallographic orientation in the electrochemical properties of silicon.
Study of the oxide film during silicon electropolishing
- An active collaboration between Palaiseau and Bath has led to a definition of identical experimental conditions in the two groups (electrolyte composition, design of a circulation cell for an appropriate control of mass-transport conditions). The infrared experiments have been performed in Palaiseau, and the ellipsometric measurements have been performed by Bath in co-operation with Southampton.
- Systematic measurements of the infrared spectra have been carried out in different electrolytes as a function of potential. The polarisation of the infrared beam has been changed, and the spectra have been analysed quantitatively. In the n-SiO spectral region, the s-spectra consist of a main line at around 1065 cm-1, corresponding to the TO component of the asymmetric stretching mode of a SiOSi group (vibration of the oxygen atom parallel to the Si-Si axis), plus two lines ascribed to defects and disorder. The p-spectra exhibit two extra lines, representing the LO counterparts of the main line and disorder mode. A quantitative analysis of these spectra has allowed to derive information on oxide thickness (from the magnitude of the signals), on oxide perfection (from the relative amount of signal associated with defects), and on oxide density (the density of Si-O vibrators is directly related to the LO-TO splitting). Oxide thickness is found to increase monotonically with potential with a more or less constant slope of 9 Å/V. This slope is somewhat larger in the region of the second current peak and near the end of the second current plateau in the typical voltammogram. The oxide thickness is found to depend little upon the electrolyte. The perfection and density of the oxides appear to be optimum near the middle of the second current plateau. This is consistent with the idea that the oxide is strongly hydrated in the first-plateau region, and possibly also near the end of the second-current plateau, where hydration is probably associated with an increase in porosity.
- In-situ ellipsometric measurements have been performed for the same electrolytes as used for the infrared measurements. The infrared and ellipsometric oxide thickness are generally in agreement within 50%. However, some deviations are present, especially at potentials more positive than 5 V vs SCE, where the thickness as derived from ellipsometry appears systematically larger. In this potential range, however, SEM indicates that substantial surface roughening occurs. One may infer those localised dielectric breakdown causes pitting and roughening of the surface, hence a rise in the dissolution rate, in the observed anodic current, and in the amount of hydrated oxide, in agreement with the infrared results. Ellipsometric measurements show that the roughening is far more severe for (100) samples than for (111) samples.
- The thickness derived from electrochemical measurements have been compared with those derived from infrared spectroscopy and ellipsometry. Coulometric measurements appear to give the correct variation of oxide thickness as a function of potential, except that the deduced values appear systematically larger by a factor of ~1.5 than those derived from infrared spectroscopy and ellipsometry. On the other hand, high-frequency capacitance measurements appear to yield an underestimated oxide thickness. The most striking fact is that the thickness derived from such measurements remains almost constant over a wide range of potential. The evident failure of the three methods to give identical values of the oxide thickness is not surprising because each method actually measures different properties of the oxide. The electrode capacitance corresponds, in principle at least, to the presence of a continuous insulating dielectric film. The results suggest that such a film is very thin and that its thickness varies only weakly with potential. Infrared spectroscopy measures the integrated intensity of the Si-O absorbency corresponding to a layer of unhydrated oxide (the dry oxide). However, it is not necessary for this layer to be continuous or insulating. A defective or porous layer of oxide is also detected. Finally, ellipsometry measures the total oxide thickness and is relatively insensitive to factors such as porosity or partial hydration.
- The defective and hydrated nature of the oxide accounts for the very high dissolution rates calculated from the current densities by assuming that the dissolution and growth rates are equal. Typically, the dissolution rate for the anodic oxides is up to two orders of magnitude higher than for the thermal oxide. This suggests that the attack of the anodic oxide by fluoride species is effectively enhanced by a large internal area, with the reaction taking place within the hydrated surface layer rather than exclusively at the surface as in the case of thermal oxide.
When a silicon electrode has been polarised for a while at anodic potential, electric charges are stored inside the interface oxide film. When the polarisation is released, some of these charges are swept back to the electrode, resulting in a transient current and a change in the interface dipole, which is experimentally accessed through flat-band potential measurements. Some other charges will disappear only upon dissolution of the oxide film. This charge decays have been modelled in Cairo. The transient currents and flat-band potentials can now be calculated and compared with experimental data. Preliminary results support the idea of a layer of positive charges close to the silicon surface, and a distribution of negative charges through the oxide. Systematic measurements are presently underway.
Reaction mechanisms in the anodic dissolution of silicon
- A systematic study of mass transport and charge-transfer kinetics in different electrolytes has been performed at Meudon, using voltammetry in a rotating-disk-electrode arrangement. Silicon dissolution appears to be limited by interface kinetics in electrolytes of low fluoride concentration, and by mass-transport in electrolytes of high fluoride concentration. The critical concentration cF* between the two regimes is of the order of 0.1 M. The values of the mass-transport-limited current are consistent with a limitation by the supply of fluoride species to the electrode.
- A striking effect of the cations present in the electrolyte has been noticed. Whilst the presence of different anions (except for F-) appears of minor importance, addition of alkali-metal ions to the electrolyte has been found to increase the anodic dissolution current, with an increasing effect upon adding heavier ions: the effect of Li+ is negligible, but an increasing current is observed in the sequence Li+
Crystallographic-orientation-dependent electrochemical effects
- A systematic study of the n-Si/fluoride interface has been undertaken at Meudon in collaboration with Constantine. The advantage of n-Si over p-Si is that the flat-band potential is determined with a better accuracy, and the effect of illumination may be studied. The effect of alkali-metal ions in the electrolyte has been investigated and (100), (111) and (110) crystallographic orientations have been studied. Preliminary results confirm that the flat-band potential and interface current are affected by both factors. Especially, these results demonstrate the possibility of determining electrode orientation from electrochemical measurements.
Follow up
- Further investigations of the electropolishing regime are in progress, using AFM and SEM for a characterisation of surface morphology. Oxide characterisation is pursued in the regime where the interface current exhibits a damped oscillating behaviour. Infrared spectra have been recorded as a function of time, during the oscillation. This gives a clear-cut indication for an oscillation of the characteristics of the oxide film (thickness, density, defectivity). The dissolution rate of the oxide film at different stages during the oscillation has also been investigated. Similar experiments are in progress using a different control of the interface (from potentiostatic to galvanostatic). The same approach is also underway using in-situ ellipsometry. In parallel, the high-frequency capacitance will be measured. We also plan experiments using microwave reflectivity to follow the changes in surface hole density (preliminary measurements have already been completed). Finally, the modelisation of the charge stored in the oxide will be tested against the measured current-time characteristics when the experiments are completed;
- Regarding the reaction mechanisms, now the mass-transport effects are well understood, and focus will be put on the kinetic aspects. The in-situ infrared experiments, which up to now have been unable to provide information on the chemical intermediates in the silicon dissolution reaction, will be pursued using a potential-modulation technique. The effect of the presence of some cations on p-Si/F- interfaces and of some dyes on n-Si/F- interfaces under illumination in aqueous electrolytes is still under investigation. Impedance of the illuminated n-Si/F- interface in non-aqueous electrolytes is presently being explored;
- Regarding the effects of crystallographic orientation, n-Si/F- interfaces of different orientations have been studied in darkness and under illumination in aqueous and in non-aqueous electrolytes. Impedance measurements have been performed and modelisation is in progress.
Study of the oxide film during silicon electropoloshing. This includes the following activities
- Determination of the oxide nature as a function of formation potential using in-situ infrared spectroscopy (Palaiseau);
- Development of electrochemical techniques for measuring the silicon oxide thickness in fluoride media during the electropolishing regimes, and calibration of these measurements using in-situ infrared spectroscopy and ellipsometric techniques (Palaiseau, Bath);
- Characterisation of the charged species incorporated into the anodic oxides, and modelling of the published experimental results of "transient flat-band potential" (Palaiseau, Cairo).
Reaction mechanisms in the anodic dissolution of silicon
- Effect of electrolyte composition upon oxide formation: role of mass-transport and charge-transfer kinetics, effect of cations upon anodic dissolution (Meudon);
- Identification of chemical intermediates in the silicon dissolution reaction, using in-situ infrared spectroscopy (Palaiseau, Bath);
- Study of silicon stabilisation by redox reagents in a fluoride electrolyte. Organic dye materials may be adsorbed or grafted on the silicon surface (Meudon, Bath, Cairo);
- Determination of the conditions for oxide control in a non-aqueous electrolyte (Constantine, Meudon).
Crystallographic-orientation-dependent electrochemical effects
- Electrochemical determination of the orientation of the electrode surface (Meudon, Constantine);
- Study of the facetting trends of the silicon surface in the electropolishing regime (Meudon, Constantine).
Topic(s)
Data not availableCall for proposal
Data not availableFunding Scheme
CSC - Cost-sharing contractsCoordinator
92195 Meudon
France